Francisella tularensis, a pleomorphic, gram-negative, facultative intracellular bacterial pathogen, is the etiologic agent of tularemia, a potentially fatal human disease. The ease with which F. tularensis can be aerosolized and its high degree of infectivity when inhaled have raised concerns about its potential for use in bioterrorism. An empirically derived, still-unlicensed vaccine strain of F. tularensis, LVS, is complicated by several issues: (i) Ft.LVS is still highly virulent in some animal models of infection, (ii) LVS vaccine has.been associated with significant undesirable side effects, (iii) Recipients of LVS vaccine develop incomplete immunity, (iv) The molecular basis for the attenuation of Ft.LVS is unknown. The mechanisms of immune protection against F. tularensis, particularly the highly virulent type A strains, are poorly defined. Our work, along with that of other investigators, has suggested that both humoral immunity (antibody to the O side chain of the lipopolysaccharide) and cellular immunity are critical for protection against this organism. Because of the incomplete understanding of immunity to F. tularensis, we have used three approaches to vaccine development: (1) screening of the vast majority of proteins in the F. tularensis proteome for potentially protective antigens; (2) marked attenuation of the live vaccine strain (Ft.LVS) through the mutation of two essential genes in the O polysaccharide (OPS) biosythesis locus, and (3) construction of a glycoconjugate vaccine composed of the full-length OPS conjugated to a carrier protein. We have obtained critical information on the nature of protective immunity and have used this information to refine our experimental strategy. Our data at this point indicate that the most effective approach will likely be some combination of our prototype vaccines. By combining the glycoconjugate vaccine with an attenuated mutant strain we have elicited protection against respiratory challenge with the wild-type type A strain Schu S4. To our knowledge, such protection has previously been reported only for Ft.LVS immunization. The combination vaccine we have developed is 7 logs less virulent in animal models than Ft.LVS; therefore, we anticipate that it will be considerably safer.